victronenergy / dynamic-ess

MIT License
76 stars 5 forks source link

[VRM - Bug]: still ilogical planning for charging/feeding/consumption #99

Closed khostri closed 8 months ago

khostri commented 8 months ago

Contact Details

jan.slaby@outlook.com

VRM portal ID

210204

Country / region

Czech Republic (cz)

B max

21

TB max

10

FB max

10

TG max

15

FG max

9.9

Battery costs

0.04

Buy price

p+0.186164629

Sell price

p-0.02214759

feed-in possible

yes

feed-in control

yes

What happened?

As reported several times algo planning is totaly wrong in my oppinion for some days. I have had DESS turned on for some time, switched from NodeRED to VRM version. I am usually ok with planning (ok I mean I can live with it but not totaly excited, as DESS is charging discharging in one hour many times, compensate for target SOC doing full power discharge of energy it was charged one minute ago etc). But planning seem totaly off today that I even switched DESS off. There is solar yield expected, battery is low on 28 percent, sell prices are low nearly on zero and system is not planning charging battery but feeding to grid for nearly zero sell price. Even from batery as soc was 28 and planned was 25... It does not make sense at all. I dont expect to plan for 100 soc (although with standard operation mode without DESS I would go for that), but not plan charging at all? Tried to leave DESS turned on and set Active soc to 85%, but in that case DESS started to feed from grid as crazy. so in the end I have turned it off in the morning.

image image

After turning it off, system operates as expected: image

When did it happen?

11/10/2023

Screenshots

No response

dirkjanfaber commented 8 months ago

I see that your system is still running 3.10. I do advise you to go to 3.11, as that does fix some things regarding the Dynamic ESS handling of the system. Mainly the quickly charging and discharging should be fixed by that.

I definitely agree that sometimes the system does things that seem illogical at first sight. If you look deeper into it, it turns out that it does make sense on how the system plans. For example, you have set your battery cost to € 0.05 / kWh (not in this report, but according to the logging of the API you do have it set to that).

So when the sun shines there are 2 possible options for the system to choose from:

Keep in mind that the second option adds € 0.10 / kWh of cost to the equation. First it costs € 0.05 / kWh for charging your battery from the solar and later it adds another € 0.05 / kWh for discharging it to the grid.

So in the above example it yields more money to discharge now (for € 0.03/kWh) instead of later today (for € 0.11 - € 0.10 = € 0.01/kWh).

There are places where the system can be improved and we are working on that. E.g. if you use more consumption then forecasted, it would be cheaper to use the €0.10 / kWh solar (from the battery) then using it from the grid.

I will close this issue as the taken choices are not considered a bug. If you want the system to take other decisions, you might want to re-evaluate the set prices in the configuration. Lowering the battery costs will increase the usage of the battery.

khostri commented 8 months ago

Hi @dirkjanfaber I understand that. My point is, that for second option you mentioned, there is not only one stream, but 2 sub options to be evaluated - your option that charging battery cost is there + discharge cost into grid. But there is also evaluation, that there is only cost for battery charge. For discharging, you are not comparing cost of discharge to grid vs sell to grid, but cost of discharging for consumption vs buying from grid to cover consumption I think. And as you see the difference of buying/selling price is high I thin that evaluation would lead to different planning :) Thanks for you support - adding such features into system is nice. I just wanted to add my points - I think that in production regular users will have similar feeling and that's why I am trying to give feedback where I see its not feeling right :)